INDEX

Explanations

greetings and thanks

The neuron fires strongly on tokens expressing polite thanks or gratitude (e.g. “Thanks,” “Thank you,” “Perfect!”), i.e. it detects polite closing/gratitude phrases.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

이면

0.62

 রক্ষণ

0.61

形式

0.56

техни

0.55

形状

0.54

 сцена

0.54

较低

0.54

িমূলক

0.53

territ

0.52

characteristics

0.52

POSITIVE LOGITS

 Thank

1.35

 thank

1.30

谢谢

1.26

 🙏

1.26

 Danke

1.26

 Gracias

1.23

 감사합니다

1.21

 gracias

1.20

 Obrigado

1.20

 teşekkür

1.19

Activations Density 0.530%